[SPARK-7943][SQL]support DataFrame created by hiveContext can create table to specific database by saveAstable(dbname.tablename)

[baishuo]

support hivecontext.sql(....).saveAsTable(dbname.tablename)

[SparkQA]

Test build #35089 has finished for PR 6868 at commit 8937597.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[baishuo]

Hi @marmbrus @yhuai ,would you please help me check whether this patch can be merged? :)

[marmbrus]

This would be a nice feature to have and thanks for working on it, but I think we need a more holistic solution than what is proposed here (there are a lot of problems with multiple databases: https://issues.apache.org/jira/browse/SPARK-8131). In particular, this issue duplicates the code that resolves a full database name into several places. Big features should probably start with a design instead of a PR.

[vidma]

notice repeated code.
IMHO relative/fully-prefixed tableName logic should be extracted into something like cannonicalTableName(relativeOrPrefixedTableName)

[marmbrus]

I believe this is fixed now and we can close this issue.